OBJECTIVE:To assess the clinical characteristics of small abdominal aortic aneurysms (AAAs) for growth patterns, growth rates, time to repair, and adverse outcomes relative to surveillance imaging intervals. PATIENTS AND METHODS:Patients with small AAAs (30-45 mm in diameter) and at least 1 follow-up imaging study were eligible for inclusion (January 1, 2014, through December 31, 2019). Variables impacting time to repair, rupture, or death were assessed. RESULTS:Of 2044 unique patients with AAAs, a random sample of 299 patients (mean ± SD age, 70.5 ± 9.0 years; 18.7% female) were divided into slow-/no- (≤2 mm per year; n=116), intermediate- (>2 to <5 mm per year; n=150), and rapid- (≥5 mm per year; n=33) growth categories. During 10.0-year median follow-up, 61.2% of the patients (n=183) underwent repair. Rapid-growth AAAs were more likely to undergo repair (88% vs intermediate 81.7% or slow/no growth 26.5%; P<.001), required earlier repair (P<.001), and were more likely to die during follow-up (79% vs 41% or 56%; P=.045). Predictors of time to repair included index AAA size, diabetes mellitus, and β-blocker or diuretic therapy. Ruptures (n=1) or impending ruptures (n=4) during follow-up were greatest in the rapid- (n=2, 6.1%; P<.001) compared with the intermediate- (n=3, 2.0%) or slow-/no- (0%) growth groups. CONCLUSION:Growth rate characteristics of AAAs influence repair-free survival, time to repair, overall survival, and rupture-free survival.

Background: Chronic venous insufficiency (CVI) can be evaluated using Duplex ultrasonography (US) and air plethysmography (PG), yet comparative performance remains unclear. Methods: We retrospectively identified patients who underwent venous insufficiency US and PG within a 90-day interval from March 1, 2015 through July 31, 2024. CVI severity was classified by the clinical (C) component of the Clinical, Etiologic, Anatomic, and Pathophysiologic (CEAP) classification system and categorized as mild (C0-2), moderate (C3), or severe (C4-6). PG and insufficiency US findings were compared across categories and machine learning models were trained to predict severe CVI. Results: We analyzed 1478 limbs from 839 patients who were predominantly women (62%) with a mean age of 61 years (± 14 years). Severe CVI was present in 32.3% of limbs. US detected venous incompetence in 67.4% of limbs overall and in 75.3% of severe CVI limbs. Abnormal PG findings occurred in 72.9% of limbs overall and in 88.5% of severe CVI limbs. PG parameters in machine learning models outperformed US parameters in predicting severe CVI (area under the receiver operating characteristic curve [AUROC] 0.82 vs 0.65). The top PG model (multilayer perceptron [MLP]) achieved an AUROC of 0.82, versus the best US model (gradient boosting) with an AUROC of 0.65. Limiting PG data to incompetence and obstruction parameters modestly reduced performance but remained higher than US (logistic regression AUROC 0.72). Conclusion: Our findings indicate that PG assessment in CVI offers superior performance compared to US. Overall, these results validate the quantitative whole-limb hemodynamic approach using PG, which provides a more complete understanding of the pathology behind CVI.

Background: Peripheral arterial disease (PAD) increases cardiovascular (CV) morbidity and mortality, but remains underdiagnosed and undertreated. Several trials support low-dose direct oral anticoagulant (DOAC) use in PAD treatment, although this has yet to be widely adopted in clinical practice. Patients and methods: We conducted a retrospective study of patients who underwent ankle-brachial index testing (ABI) from 1996 - 2020 at Mayo Clinic. We included patients with PAD defined by abnormal ABI (<1.0 or >/=1.4). Primary outcomes evaluated were myocardial infarcts (MI), ischemic strokes (IS), critical limb ischemia (CLI)/amputation, bleeding events and all-cause mortality. DOAC and warfarin use were each compared to no anticoagulant use for the outcomes using univariate analysis and multivariate analysis. Results: 22,162 patients had abnormal ABI readings; 1,266 were on warfarin and 269 were on DOAC for any indication. Both the DOAC and warfarin groups showed significant a decrease in all-cause mortality. The DOAC group showed superior mortality outcomes with HR 0.50 [95% CI 0.40-0.63], p-value <0.001 compared to warfarin with HR 0.88 [95% CI 0.81-0.96], p-value <0.004. There appeared to be a similar trend for MI and CLI/amputation however this was not statistically significant. IS was similar with only warfarin being statistically significant. The DOAC group had improved bleeding outcomes compared to the warfarin group, HR 0.53 (95% CI 0.24-0.85), p-value 0.007. Notably, the addition of ASA for both AC groups resulted in significant HR >1. Conclusions: Our study shows that anticoagulation use, particularly DOACs, is associated with decreased all-cause mortality in patients with PAD. There appears to be a favorable trend for DOACs in MI, IS and CLI/amputation. Lastly, DOACs were found to have superior outcomes with bleeding events.

Ankle brachial index (ABI) can be unreliable in patients with non-compressible vessels. Our aim is to determine the feasibility of toe brachial index (TBI) and reporting criteria in a large population. We evaluated Doppler waveforms and segmental pressures in 26,719 limbs. TBI was obtained in 92.7%, mean TBI = 0.61 ± 0.25. TBI was obtained in 82%of limbs with unobtainable ABI. In hemodynamically normal subgroup (defined as those with normal ankle-brachial indices at rest and after exercise) the mean TBI was 0.84 ± 0.14. In severe PAD subgroup (defined as ABI < 0.5 and monophasic waveforms) the mean TBI was 0.16 ± 0.12. Limbs with a diagnosis of a PAD (ABI ≤ 0.9) had a TBI <0.8 in 99.5% of the cases, and <0.6 in 90% of the cases. A TBI of 0.8 had a negative predictive value for PAD of 0.99. A TBI cutoff of 0.6 had a positive predictive value for PAD of 0.95. Based on these results we propose defining normal TBI above 0.8, borderline between 0.8 and 0.61, abnormal TBI ≤ 0.6 and severe PAD as TBI ≤ 0.2. In conclusion TBI can be reliably measured in patients with PAD and offer valuable information when diagnosing PAD. We present our diagnostic criteria based on clinical data.

Warfarin is a commonly prescribed anticoagulant with a narrow therapeutic window, which requires frequent and specialized monitoring. This work aims to develop standardized optimal warfarin dose decision support using a machine learning model based on time-series anticoagulation data and patient demographic characteristics. We propose an offline reinforcement learning model (RL) using a Batch-Constrained Q-Learning algorithm (BCQ) in a discrete action setting to predict the cumulative warfarin dose for the days until the next INR (International Normalized Ratio) test. Prior approaches utilized time-series supervised learning methods such as regression or Long Short Term Memory (LSTM) neural networks. The key advantage of reinforcement learning is its capacity to learn optimal dosing strategies from suboptimal clinical states in the data. To evaluate the model we compared the predicted warfarin doses with the physician-prescribed doses. Our BCQ model with a prediction accuracy of 98.6% significantly outperformed our baseline Long Short Term Memory (LSTM) model with a prediction accuracy of 71.09% . Further qualitative evaluation for explainability indicated that the model correctly adjusted the warfarin dose at time steps when patients had out-of-range INRs.

Background: Whether oral anticoagulation (OAC) can be discontinued after the resolution of left ventricular thrombus (LVT) remains controversial, particularly in patients without risk factors such as left ventricular aneurysms. This study aimed to identify patients who may safely discontinue OAC. Methods: We conducted a retrospective cohort study of patients with LVT resolution who completed their anticoagulation course and excluded patients with left ventricular (LV) aneurysm. Clinical and echocardiographic factors were compared between patients who experienced LVT recurrence (n = 22) and those without recurrence during follow-up (n = 94). Results: The cohort had a mean age of 64.5 ± 15.0 years, and half (50%) were male. Most had an ischemic cardiomyopathy (64.6%). LVT recurred in 19% (22/116) of patients following initial resolution. Patients with recurrence were older (mean age 71.3 ± 11.9 vs. 63.4 ± 15.7 years, p = 0.03) and were more likely to have prior ischemic stroke and arterial thromboembolism (18.2% vs. 4.3%, p = 0.02 for both). At the time of thrombus resolution, LVEF was significantly lower among patients who had recurrence (33.4 ± 13.6% vs. 42.5 ± 15.5%, p = 0.012). Stroke, systemic embolism, and mortality rates were similar regardless of recurrence status. Recurrence risk by presence or absence of stroke and LEVF >50% on the LVT resolution echocardiogram is shown in Figure 1. Notably, no recurrences were observed among patients with both LVEF > 50% and no history of stroke (0/32), which was statistically significant compared to each other category. Conclusion: Among patients without left ventricular aneurysm who discontinued oral anticoagulation after LVT resolution, those with preserved LVEF (>50%) and no prior history of stroke had a very low risk of recurrence, with no events observed in this subgroup. These findings suggest that some patients may safely discontinue anticoagulation after LVT resolution.

BACKGROUND:The calf muscle pump is an understudied contributor to venous return from the lower extremity. This study aimed to determine if calf pump function (CPF) is independently associated with the severity of chronic venous disease classified by CEAP (Clinical-Etiology-Anatomy-Pathophysiology). METHODS:The Mayo Clinic Vascular Laboratory database was analyzed from January 2015 through September 2023. Ambulatory adults who underwent venous air plethysmography were included. Venous plethysmography assessed the severity of venous incompetence, and CPF was measured as ejection fraction (EF) per leg. The clinical component (C0 through C6) of the CEAP score was evaluated for each extremity at the time of the study. RESULTS:A total of 7760 limbs from 3733 patients were analyzed. The mean age was 62 years, with 62% women. Venous obstruction was detected in 3.2% of limbs. Venous incompetence severity was categorized as normal (44%), mild (26%), moderate (19%), and severe (10%). A significant trend of reduced CPF was observed with higher CEAP scores (p < 0.001). Multivariable logistic regression, adjusted for age, sex, degree of venous incompetence, and obstruction showed reduced CPF was a significant predictor (odds ratio 1.84, CI: 1.5-2.2) of active/prior ulcer (C5 or C6). In contrast to more severely reduced CPF, mildly reduced CPF (EF 40-49%) was not associated with active/prior ulcers. CONCLUSION:This large contemporary study demonstrates that decreased CPF is associated with worse chronic venous disease. Importantly, we demonstrate for the first time that CPF is independently associated with active/prior venous ulcers after accounting for other venous physiologic parameters and demographics.

BACKGROUND:Chronic venous insufficiency (CVI) results in complications such as pain, swelling, edema, skin changes, and ulcerations of the lower extremities. Valvular incompetence and venous obstruction are well-recognized contributors to CVI. Post-exercise venous refilling time (P-EVRT), the time to refill veins after calf muscle contractions during venous plethysmography study, is an understudied contributor to CVI. METHODS:In this cross-sectional study of 4755 patients who were evaluated with venous air plethysmography, 9510 lower limbs were categorized based on P-EVRT into two groups: rapid (<20 seconds; n = 5256) and normal (n = 4254). RESULTS:Rapid P-EVRT was associated with higher mean CEAP scores (3.2 vs 2.5; P < .001) and a higher prevalence of active/prior ulcers (11.6 vs 4.1%; P < .001). Univariable analysis showed that age, male sex, the severity of incompetence, obstruction, calf pump function, and rapid P-EVRT were all significantly associated with active/prior ulcers. After multivariable adjustment for these significant factors, rapid P-EVRT was an independent contributor to active/prior ulcers (odds ratio, 1.44; 95% confidence interval, 1.17-1.77). Among limbs without other venous pathology by plethysmography (incompetence, obstruction, reduced calf pump function), rapid P-EVRT remained significantly associated with higher mean CEAP scores (P < .001) and a higher prevalence of venous ulcers than limbs with normal P-EVRT (5.7% vs 2.7%; P = .001). CONCLUSIONS:In this large contemporary study using venous air plethysmography, we demonstrate that rapid P-EVRT is an important and unique venous physiologic parameter that informs our understanding of the clinical severity of CVI.

Background: Exercise stress testing uses metabolic equivalents of tasks (METs) to measure the energy cost of activities, aiding in the assessment of exercise capacity and cardiovascular health. Despite its significance, the correlation between calf muscle pump function (CPF) and exercise stress testing remains unexplored. We aimed to evaluate the relationship between CPF and peak METs as determined by cardiopulmonary treadmill exercise stress testing. Methods: The study included adults who underwent exercise cardiopulmonary stress testing and venous plethysmography at Mayo Clinic between April 2017 and March 2020. The protocols other than Bruce, Mayo, Modified Naughton, and Naughton protocols were excluded. The CPF ejection fraction (EF) was calculated per leg based on refill volumes post-exercise as a percentage of passive drain refill. The classification of CEAP (Clinical-Etiology-Anatomy-Pathophysiology) was utilized to better understand chronic venous insufficiency (CVI). Results: A total of 155 patients who underwent both exercise stress testing and venous plethysmography were included, with a mean age of 61.31 ± 14.03 years, and 84 (54.2%) were male. The peak measured METs for normal, unilaterally reduced, and bilaterally reduced CPF were 8.5 (2.5), 7.3 (2.1), and 7.1 (2.4), respectively (p=0.004, Figure 1). Multiple linear regression models were developed with METs as the outcome to determine if CPF was an independent predictor of METs on cardiopulmonary exercise stress testing. IIn model 1, the following independent variables were included: resting heart rate, peak heart rate, peak systolic blood pressure, recovery heart rate at minute 1, and worst EF (Table 1). In model 1, with only exercise parameters, lower EF was associated with lower METs (p=0.03). In a second analysis, variables identified as statistically significant with METs in the initial model were included, along with CEAP class (model 2) and CCI (model 3) (Table 2). In model 2, CEAP class 3 or higher was associated with decreased METs on the exercise stress test. This correlation implies that individuals with moderate to severe CVI may influence exercise capacity, demonstrating the interconnectedness of the cardiovascular system. Moreover, in model 3, the CCI, a predictor for mortality, was not significantly associated with METs. Conclusion: Our findings revealed that more severe CVI (CEAP class and reduced CPF) was associated with reduced exercise capacity after accounting for other factors.

ObjectiveTo evaluate mortality outcomes by varying degrees of reduced calf muscle pump (CMP) ejection fraction (EF).Patients and MethodsConsecutive adult patients who underwent venous air plethysmography testing at the Mayo Clinic Gonda Vascular Laboratory (January 1, 2012, through December 31, 2022) were divided into groups based on CMP EF for the assessment of all-cause mortality. Other venous physiology included measures of valvular incompetence and clinical venous disease (CEAP [clinical presentation, etiology, anatomy, and pathophysiology] score). Mortality rates were calculated using the Kaplan-Meier method.ResultsDuring the study, 5913 patients met the inclusion criteria. During 2.84-year median follow-up, there were 431 deaths. Mortality rates increased with decreasing CMP EF. Compared with EF of 50% or higher, the hazard ratios (95% CIs) for mortality were as follows: EF of 40% to 49%, 1.4 (1.0 to 2.0); EF of 30% to 39%, 1.6 (1.2 to 2.4); EF of 20% to 29%, 1.7 (1.2 to 2.4); EF of 10% to 19%, 2.4 (1.7 to 3.3) (log-rank P≤.001). Although measures of venous valvular incompetence did not independently predict outcomes, venous disease severity assessed by CEAP score was predictive. After adjusting for several clinical covariates, both CMP EF and clinical venous disease severity assessed by CEAP score remained independent predictors of mortality.ConclusionMortality rates are higher in patients with reduced CMP EF and seem to increase with each 10% decrement in CMP EF. The mortality mechanism does not seem to be impacted by venous valvular incompetence and may represent variables intrinsic to muscular physiology.

Background Patients with peripheral artery disease are at increased risk for major adverse cardiac events, major adverse limb events, and all‐cause death. Developing tools capable of identifying those patients with peripheral artery disease at greatest risk for major adverse events is the first step for outcome prevention. This study aimed to determine whether computer‐assisted analysis of a resting Doppler waveform using deep neural networks can accurately identify patients with peripheral artery disease at greatest risk for adverse outcome events. Methods and Results Consecutive patients (April 1, 2015, to December 31, 2020) undergoing ankle–brachial index testing were included. Patients were randomly allocated to training, validation, and testing subsets (60%/20%/20%). Deep neural networks were trained on resting posterior tibial arterial Doppler waveforms to predict major adverse cardiac events, major adverse limb events, and all‐cause death at 5 years. Patients were then analyzed in groups based on the quartiles of each prediction score in the training set. Among 11 384 total patients, 10 437 patients met study inclusion criteria (mean age, 65.8±14.8 years; 40.6% women). The test subset included 2084 patients. During 5 years of follow‐up, there were 447 deaths, 585 major adverse cardiac events, and 161 MALE events. After adjusting for age, sex, and Charlson comorbidity index, deep neural network analysis of the posterior tibial artery waveform provided independent prediction of death (hazard ratio [HR], 2.44 [95% CI, 1.78–3.34]), major adverse cardiac events (HR, 1.97 [95% CI, 1.49–2.61]), and major adverse limb events (HR, 11.03 [95% CI, 5.43–22.39]) at 5 years. Conclusions An artificial intelligence–enabled analysis of Doppler arterial waveforms enables identification of major adverse outcomes among patients with peripheral artery disease, which may promote early adoption and adherence of risk factor modification.

OBJECTIVE:Reduced calf muscle pump function (CPF) is an independent risk factor for venous thromboembolism and mortality. We aimed to evaluate the relationship between handgrip strength (HGS) and CPF. METHODS:Patients referred to the Gonda Vascular Laboratory for noninvasive venous studies were identified and consented. Patients underwent standard venous air plethysmography protocol. CPF (ejection fraction) was measured in each lower extremity of ambulatory patients by comparing refill volume after ankle flexes and passive refill volumes. The cutoff for reduced CPF (rCPF) was defined as an ejection fraction of <45%. Maximum HGS bilaterally was obtained (three trials per hand) using a dynamometer. HGS and CPF were compared (right hand to calf, left hand to calf) and the correlation between the measures was evaluated. RESULTS:115 patients (mean age, 59.2 ± 17.4 years; 67 females, mean body mass index, 30.83 ± 6.46) were consented and assessed for HGS and CPF. rCPF was observed in 53 right legs (46%) and 67 left legs (58%). CPF was reduced bilaterally in 45 (39%) and unilaterally in 30 (26%) patients. HGS was reduced bilaterally in 74 (64.3%), unilaterally in 23 (20%), and normal in 18 (15.7%) patients. Comparing each hand/calf pair, no significant correlations were seen between HGS and CPF. The Spearman's rank correlation coefficients test yielded values of 0.16 for the right side and 0.10 for the left side. CONCLUSIONS:There is no significant correlation between HGS and CPF, demonstrating that HGS measurements are not an acceptable surrogate for rCPF, indicating different pathophysiological mechanisms for each process.

The calf muscle pump is a robust venous pump that is a major contributor to blood return from the lower extremities. Recent studies have found reduced calf pump function (CPF) to be a predictor of venous thromboembolism (VTE) and mortality.1,2 Measurement of mortality and VTE risk in prior studies used previously established clinical cut-off values to classify normal versus reduced CPF. In this study, we sought to re-evaluate quantitative plethysmography measurements (passive outflow, passive refill, and postexercise refill) to determine if these previously established thresholds of CPF could be refined when it relates to prediction of deep vein thrombosis (DVT). To accomplish this, we used long-term outcome data from the Rochester Epidemiology Project and examined Olmsted County residents who underwent clinical venous plethysmography testing between 1998 and 2015. Patients were excluded if they had a history of VTE or if there was evidence of venous outflow obstruction on the index plethysmography study; see previously published methods.1 CPF measurements were analyzed as refill volumes as a percent of leg passive drain refill (PDR) volume (standard) or leg outflow volume, or using the crude volume refilled after exercise (toelifts), regardless of testing methodology (strain gauge or air). In addition to plethysmography data, venous continuous wave Doppler assessment was performed/supervised by Registered Vascular Technologists at each plethysmography study and the presence of and location of deep venous insufficiency was analyzed in conjunction with data on overall venous hemodynamic insufficiency. The venous plethysmography parameters of each examined leg were evaluated with the outcome of interest being imagingor autopsy-confirmed ipsilateral DVT (proximal or distal) after the index study using Cox proportional hazard models. The Charlson Comorbidity Index (includes: myocardial infarction, congestive heart failure, peripheral vascular disease, stroke or transient ischemic attack, dementia, chronic obstructive pulmonary disease, connective tissue diseases, peptic ulcer disease, liver diseases, renal diseases, diabetes mellitus, hemiplegia, chronic kidney disease, cancers, and HIV/AIDS) as well as age, body mass index, and sex were evaluated initially in univariates analyses. The research was reviewed and approved by Institutional Review Boards from Olmsted Medical Center and Mayo Clinic in Rochester, MN. A total of 3064 legs (1532 patients) were analyzed with a median follow-up of 11.7 years. The median age of the cohort was 63.9 years (SD 18.4) and the majority were female (69%). CPF was reduced by previously established criteria bilaterally in 38%, unilaterally in 23%, and was normal bilaterally in 39%. Venous insufficiency by plethysmography was present (mild, moderate, or severe) in 52.6% of right legs and 55% of left legs (p = 0.11). The primary outcome of ipsilateral leg DVT occurred in 1.8% of right legs and 2.8% of left legs (p = 0.053). We determined the optimal predictive cut-offs for CPF using continuous variables for each calculation method. Each method for calculation of CPF was highly associated with development of ipsilateral leg DVT. However, the parameter that was most predictive was the standard calculation of CPF using passive refill volumes in comparison to exercise refill Reduced calf pump function and proximal deep vein incompetence are predictors for ipsilateral deep vein thrombosis

Background: Peripheral artery disease (PAD) is a risk factor for adverse limb events (LE) and cardiovascular events (CVE) that coexists with type 1 (T1) and 2 (T2) diabetes mellitus (DM). Little is known about comparative risk of LE and CVE in T1/T2 DM patients with PAD. Patients and methods: We queried our database of 40,144 patients ≥18 years old who underwent ankle brachial index (ABI) measurement from 01/1996-02/2020. We isolated T1/T2 DM patients with PAD diagnosed by ankle brachial index (ABI; low [<1.0] or elevated [>1.4]) and retrieved demographics including glycated hemoglobin (HbA1c). Primary outcomes were LE (critical limb ischemia/vascular amputation) and CVE (myocardial infarction/ischemic stroke). All-cause mortality was a secondary outcome. Multivariable Cox proportional regression yielded hazard ratios (HR) with 95% confidence intervals (CI) after adjusting for pertinent risk factors including age, hypertension, hyperlipidemia, smoking, and HbA1c. Results: Our study found 10,156 patients with PAD and DM (34% T1DM, 66% T2DM) with median follow-up time 34 mo (IQR 85 mo). T1DM patients were younger than T2DM (mean age 67 vs. 70 years), with higher median HbA1c (7.7 [IQR 1.9] vs. 6.7% [IQR 1.6]), and more prevalent hypertension, hyperlipidemia, CAD, and CKD. Antiplatelet and statin use was equivocal. Elevated ABI was more common in T1DM (47 vs. 28%). LE occurred in 23% and CVE in 12% patients. LE risk was higher in T1 than T2 DM patients (HR 1.58 [95% CI 1.44, 1.73], p<0.0001), but CVE and all-cause mortality were equivocal. These observations were preserved across ABI and HbA1c subgroup analyses. Conclusions: PAD patients with T1DM had a higher LE risk than those with T2DM, even after adjustment for glycemic control and pertinent risk factors, but CVE risk and all-cause mortality were equivocal. These data suggest a potential role for more intensive LE risk modification in PAD patients with T1DM, but further investigation is needed.

Introduction: Predictive algorithms using multiple clinical variables can identify patients at higher risk for abdominal aortic aneurysms (AAA), however they are cumbersome and cannot used directly by patients. A simple, cheap, noninvasive, and readily available (virtual) screening tool for identification of AAA could aid in screening algorithms. Aims: We sought to determine if a machine learning algorithm could predict the presence of an AAA by using a image of the face Methods: Diagnostic imaging studies were extracted from the electronic health record (EHR) from 5/7/2018 through 1/1/2023 and analyzed by a natural language processing algorithm previously validated to identify abdominal aortic aneurysms. Patient EHR profile pictures were extracted and matched to imaging studies. Various deep neural network (DNN) architectures were explored, all trained as classifiers on the face images along with clinical variables to predict clinical outcomes. All models used were CNNs with an EfficientNet architecture. Model performance was evaluated by standard metrics such as the area under the curve (AUC), specificities (Sp), and sensitivities (Sn). All performance metrics reported results from the test subset of the primary dataset and did not contain any observations considered during training. Results: A total of 5522 patients with facial images and AAA diagnostic studies were analyzed, among whom 1314 (23.8%) had a AAA. The optimal model had an AUC of 0.72 (Sn 0.44, Sp 0.83) for predicting AAA and remained similar in women (AUC 0.7), men (AUC 0.65), age >65 (AUC 0.63), and age <=65 (AUC 0.72) (Table). Conclusions: Using only an image of a patients face, the presence of an AAA could be identified with moderate performance in men, women, and in those less than 65. The model’s performance in demographics not typically screened for AAA may allow for targeted screening programs in these groups.

Introduction: Warfarin is a commonly prescribed anticoagulant for treating atrial fibrillation, mechanical valves, and venous thromboembolism. Warfarin dose management remains challenging due to dosing variability between patients and warfarin’s narrow therapeutic window. Time in therapeutic range (TTR) is critical to warfarin’s safety and efficacy, but TTR typically remains low (40-50%) in community practices. Specialized anticoagulation clinics and protocolized approaches can increase TTR but have great administrative burdens and health care costs. Aim: To develop standardized optimal warfarin dose decision support using a machine learning model based on time series anticoagulation data and patient demographic characteristics Methods: The dataset included 12,497 warfarin patients monitored in the anticoagulation tracker of electronic medical records across the Mayo Clinic Enterprise. The dataset contained time series anticoagulation data (warfarin dose, INR, INR target range) and patient demographic characteristics. We implemented an offline deep reinforcement learning model (DRL) to predict the cumulative warfarin dose for the days until the next INR test based on a patient’s historical anticoagulation data. Prior approaches utilized traditional supervised learning methods such as regression or Long Short Term Memory (LSTM) neural networks to approximate a function mimicking the behavior of the training data. On the other hand, DRL learned an optimal dosing policy through continuous interaction and feedback from the training data. The key advantage of DRL is the model can learn to behave differently (and potentially better) for suboptimal clinical states in the data such as overdosing or underdosing. To evaluate the DRL model we compared the predicted warfarin doses with the physician-prescribed doses Results: DRL model’s prediction accuracy was 96.96%, outperforming our implementation of a baseline LSTM model with a prediction accuracy of 70.58%. Further evaluation of the DRL model indicated that the model correctly adjusted the warfarin dose at time steps when patients had out-of-range INRs. Conclusions: Offline deep reinforcement learning demonstrates potential in supporting warfarin dose management to maximize TTR.

Fast-growing abdominal aortic aneurysms (AAA) have a high rupture risk and poor outcomes if not promptly identified and treated. Our primary objective is to improve the differentiation of small AAAs' growth status (fast versus slow-growing) through a combination of patient health information, computational hemodynamics, geometric analysis, and artificial intelligence. 3D computed tomography angiography (CTA) data available for 70 patients diagnosed with AAAs with known growth status were used to conduct geometric and hemodynamic analyses. Differences among ten metrics (out of ninety metrics) were statistically significant discriminators between fast and slow-growing groups. Using a support vector machine (SVM) classifier, the area under receiving operating curve (AUROC) and total accuracy of our best predictive model for differentiation of AAAs' growth status were 0.86 and 77.50%, respectively. In summary, the proposed analytics has the potential to differentiate fast from slow-growing AAAs, helping guide resource allocation for the management of patients with AAAs.